Recommendations for Inpatient Pediatric CPOE/CDS Systems - A Systematic Review

,

The whole process was reviewed and discussed regularly by an in-house group of six clinical pharmacists and a clinical pharmacologist experienced in pharmacoepidemiology, drug therapy, and e-prescribing systems.

Definitions
A computerized physician order entry (CPOE) system is defined as a software application that allows physicians to prescribe medications electronically and ensures standardized, legible, and complete medication orders [8,9].
A clinical decision support (CDS) system is defined as a computerbased system that is able to integrate and automatically evaluate patient-specific medication and laboratory data and thereby provide decision support for the user [ 8,10].
A feature was defined as a task within a CPOE/CDS system.Features allow the CPOE-CDS system to run calculations, interface with external software systems, and output information/results to the user or to another part of the system.In addition a feature enables the user to interact with the CPOE/CDS system.

Search Strategy
The systematic review was conducted using the MEDLINE (via PubMed) and EMBASE (via the Deutsches Institut für Medizinische Dokumentation und Information [DIMDI]) databases in July 2013 with no restrictions on publication date, language, or study type.Non-scientific literature was not included in the review.
Tables 1-A and 1-B show the MEDLINE and EMBASE search queries used to conduct the search.We had only access to the titles of the identified records in EMABSE.Therefore the corresponding abstracts and full-text articles were searched in PubMed and Google Scholar.In accordance with the Cochrane Handbook for Systematic Reviews and the publication of Smith at al., we chose both databases to conduct a comprehensive systematic review of the majority of the recent records [7,11].The search strategies were determined at by modifying a strategy developed by van Rosse et al [9].

Study Selection
In order to identify eligible studies for the systematic review, we set up a predefined list of inclusion and exclusion criteria.Titles, abstracts, and full text articles were included if they reported about recommended CPOE +/-CDS features important to pharmacological therapy in general pediatric hospital settings and were published within the last 10 years.Articles were excluded (I) if they contained recommendations for adults (patients ≥18 years of age), oncology, cost, surgery, infusion, diagnostic support, or non-pharmacological setting; or (II) if an EMBASE record could not be clearly allocated to its full-text article due to title equivalents in PubMed and Google scholar search.To avoid a wrong selection of an article, those EMBASE records were excluded.

Introduction
Over the past decade in the USA and recently in the EU programs have been initiated to foster electronic-health (e-Health) and systems for the electronic prescription of drugs [1,2].Particularly in the USA, these programs lead to a significant increase in the use of computerized physician order entry (CPOE) systems with clinical decision support (CDS) in pediatric care [3].Between the years 2006 and 2011, the percentage of children's hospitals that use CPOE systems with integrated CDS rose from 6% to 59% [3,4].
The increasing use of technologies and methods in health care usually leads to guidelines and standards due to harmonization processes.One of the most famous examples was the "International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use" (ICH).It set up guidelines for drug registration and standards for the used terminology [5].These are intended to simplify information transfer, give guidance to involved persons and public authorities and, finally, improve public health.Systematic overviews of present standards and recommendations lay the groundwork for harmonization processes.Until now, there have been no systematic reviews about recommended CPOE/CDS features for pediatric pharmacotherapy from literature published worldwide.

Objective
Our objective was to systematically review the literature for descriptions of features actually recommended for pediatric CPOE/ CDS systems in hospital settings worldwide.

Methods
To accomplish our objective, we conducted a systematic review of the literature describing pediatric CPOE/CDS features recommended for hospital settings.
A systematic search and screening of articles for eligibility based on a predefined protocol that incorporated keywords as well as inclusion and exclusion criteria was followed by extraction and analysis of data.The methodology for the search was specified beforehand and documented in a protocol according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) Statement and as described by Smith et al [6,7].features closely aligned with the latest policy statement of the AAP.
The simplification process is illustrated below using two recommended features of the AAP: "drug-drug interaction checking" and "indication-based dosing and individual and daily dose alerts using a mg/kg per day or mg/m 2 per day formula" [12].
The term "drug-drug interaction checking" was renamed to "drugdrug interaction check" and used as a standardized feature.The second recommended term was quite long and was therefore split into 7 standardized features as follows: indication, single-dose range check and alerts, daily-dose range check and alerts, dosing by weight, dosing by body surface area, body surface area documentation, and weight documentation.The body surface area and weight documentation are not part of the AAP-recommended features but were deduced from its context and added to our standardized set.This approach was judged to be meaningful because a dose cannot be properly calculated without the inclusion of the underlying patient information.

Grading of Standardized Recommended Features for CPOE/CDS
The American Academy of Pediatrics graded the recommended features within their latest statement according to "minimum, " "when possible" and "could be" requirements [12].For simplicity, these grades were transformed into the more visually intuitive symbols of (+++) for "minimum requirements, " (++) for "when possible requirements" and (+) for "could be" requirements, and then applied to the standardized recommended features identified [12].Finally, the graded features were allocated to the following sections: CPOE, CDS, external databases or hospital information system.Thereby, we differentiated between the typical sections of a Based on these eligibility criteria, titles and abstracts of the researched articles were screened and duplicates were removed.The remaining full text articles were read and assessed for eligibility.The eligibility assessment was conducted cooperatively in a standardized manner by one author and an internship pharmacist.Disagreements were resolved by re-assessing the inclusion and exclusion criteria individually as they applied to the articles in question, followed by discussion until consensus was achieved regarding their inclusion or exclusion.

Data Extraction
The selected papers were extracted according to useful terms of recommended features, the source (e.g., the American Academy of Pediatrics [AAP]), the type of information (e.g., policy statements), the year of publication, and the country where the research was performed.

Standardization of recommended Terms
During the data extraction it became obvious that no unified terminology has yet been established to describe the features of CPOE/CDS systems.Therefore, we performed a standardization of such terminology.As a basis for this exercise, we followed the proposed terms of the latest policy statement of the AAP and added terms from the other selected publications if they were deemed to be useful [12].To simplify the language, terms that were too long or too complicated were modified by renaming, shortening and splitting.This standardization was intensively discussed by the two authors and disagreements were resolved by re-assessing and discussing each term in question until consensus was achieved.
In the end, we generated a standardized set of recommended Beside the performed grading and mentioned classifications, we were able to define six "pediatric-exclusive" features, which were primarily used for pediatric patients: "automatic strength to volume conversion, " "pediatric specific adverse effect alerts, " "choose suitable pediatric drug formulations, " "off-label treatment protocols, " "pediatric specific medication catalogs, " and "choosing extemporaneous formulations."

Study Selection
A total of 1057 records could be identified and screened through the study selection process.Of these, 15 full-text articles were assessed for eligibility through the title and abstract screening process, which resulted in 5 studies being considered as having met the inclusion criteria.These studies were included in the data extraction procedure (Figure 1).
Of these 42 recommended features, 34 met the definition of a CPOE or CDS feature (14 CPOE features and 20 CDS features).Out of the remaining 8 features, 2 were allocated to the area "external database" and 4 were categorized as "hospital information system" features.The last 2 features (connection to vendor [+] and connection to pharmacy [+]) could not be allocated to any of the sections mentioned above.

Recommended features of pediatric CPOE/CDS systems
The 16 minimum requirements for pediatric CPOE/CDS systems

Study Characteristics
The five articles describing recommendations for CPOE and CDS features were published between 2007 and 2013 [12][13][14][15][16].They included policy statements, technical reports, a combination of a survey and a practice report, or contained only recommendations for CPOE/CDS features.Out of these five articles, four were published in the USA and one in the Netherlands.The American Academy of Pediatrics published three of the articles.

Standardization and Grading of Identified Features
From the five included articles, we extracted 68 terms that were then simplified and standardized to 42 features.Sixteen of those were graded as "minimum" requirements (+++), 19 as "when possible" requirements (++), and 7 as "could be" requirements (+) as shown in Table 2.
comprised basic dosing information as well as basic anthropometric information.Foremost among these were weight documentation and dose entry as CPOE features, and the highly important features "doing by weight" and single-and daily-dose range checks and alerts, as CDS features.Additionally, two "pediatric exclusive" features -"pediatric specific medication catalogs" and "automatic strength to volume conversion" -were classified as minimum requirements of pediatric CPOE/CDS systems.
The majority of the 19 identified "when possible" requirements was categorized as CDS features.These requirements recommended the use of drug-drug, -allergy, and -food interaction checks as well as order sets in pediatric CPOE/CDS systems.In addition, we identified three "pediatric exclusive" features -"the pediatric specific adverse effect alerts", "choosing extemporaneous formulations" and "off-label treatment protocols." Out of the seven "could be" requirements five fit the definition of a CDS function.These contained the display of percentile charts, laboratory and duplicate checks, dosing by age, and the "pediatric exclusive" feature "choose suitable pediatric drug formulations."

Discussion
In this article, we systematically reviewed literature published worldwide.We wanted to investigate which features are actually recommended for pediatric CPOE/CDS systems in the published literature.As a result of this review, 42 standardized recommended features were identified, based mainly on publications of the American Academy of Pediatrics.Based on information gathered from the AAP, these were divided into 16 "minimum, " 19 "when possible" and seven "could be" recommendations.This is the first systematic review summarizing the current recommendations regarding features essential in pediatric CPOE/ CDS systems -primarily based on statements of the AAPpublished globally.Of the 42 identified recommended features, 34 were contained in the publications of the AAP, providing a clear indication of the features important for pediatric e-prescribing  [15,16] Because of the differences in the quality of evidence in these reports relative to that in the statements of the AAP, their recommendations have been graded as being less important.However, the non-AAP articles provide further information on recommendations exclusively specific to pediatric patients (e.g., recommendations related to off-label treatment protocols and the choice of suitable pediatric drug formulations), and are therefore highly important in that they add important information to the overall picture of recommended features for these systems.
The grades of the recommended features identified in the systematic review correspond with the grades of the recommended features of the "Children Electronic Health Record Formats" (CEHRF).
A review of the CEHRF showed that out of approximately 700 features or normative statements of the CEHRF only 48 refer to CPOE or CDS features and grades.The recommendations of the CEHRF were classified in three grades: shall, should and can [17,18].Thirty-seven these features or normative statements could be assigned to one of 16 recommended features of the systematic review; 75% of the 37 CEHRF features had the same grade of recommendation as the corresponding features in our systematic review.An explanation for the small number of corresponding features between the CEHRF and the systematic review is that CPOE/CDS features are only a small part of an electronic health record [13].In addition, due to the general description of the features in the CEHRF, only a few points of correspondence between the CEHRF and the systematic review could be found.
The geographical distribution of the articles included in this review suggests a delay in research about country-specific requirements for pediatric CPOE/CDS systems within the European Union

Features of external databases
Provision of drug information Access to evidence database NS to promote, among other actions, the meaningful use of EHRs and imposes penalties if no EHR is implemented by 2015; as a result, the use of EHRs doubled in the first 2 years after the law went into effect [2,19,20].In 2012, the EU Commission published an action paper to promote eHealth technologies in the EU [1].One reason for this action plan was the insight that the EU lags at least 10 years behind in the implementation of IT solutions in healthcare [1].

NS
The grading of the identified recommendations is well in line with its clinical relevance.The identified minimum requirements provide all important features to enable complete and appropriate prescriptions."Weight documentation" and "dosing by weight" are one of the most important features in pediatric care and therefore shall be a part of any pediatric CPOE/CDS system.The "when possible" requirements add medication safety aspects to pediatric CPOE/CDS systems, like drug-drug or drug-allergy interaction checks.It is already known that the implementation of features like 'drug-drug interaction check' or 'drug-allergy check' is able to eliminate up to 100% of the associated medication errors [21,22].However, medication errors due to non-detection of drug-drug interactions or drug-related allergies have less severe consequences than errors resulting from incorrect dose calculations [21,23].Therefore a "when possible" classification for those features is reasonable.The "could be" requirements comprise various types of features like "dosing by age".It is an important feature for dosing assessment of drugs like paracetamol in neonates.However, the high variation of body weight within age groups generally outweighs the importance of an age based dosing support in as compared to a weight based dosing.Therefore, the classification of this feature as a "could be" requirement seems to be reasonable.
Our results indicated that "pediatric-exclusive" features were rated from the AAP as less important compared to the "standard" features of pediatric e-prescribing systems."Pediatric specific medication catalogs" and "automatic strength to volume conversion" were classified as "minimum requirement" of pediatric CPOE/CDS systems.The recommendations for "pediatric specific adverse effect alerts, " "choosing extemporaneous formulations, " and "offlabel treatment protocols" were classified as "when possible" and "choose suitable pediatric drug formulation" as "could be." In general, these six "pediatric exclusive" features are associated with the characteristic considerations of pediatric clinical care, such as off-label use [24][25][26], the use of liquid dosage or suitable pediatric formulations [27], and extemporaneous formulations, as well as the consideration of variations in adverse effects in pediatric inpatients (e.g., Reye syndrome).It is already known that children are more vulnerable to adverse drug events, especially those caused by errors that could be avoided if these features were available in CPOE/CDS systems [24][25][26][28][29][30].Given the possible benefits in medication safety due to the implementation of these features in pediatric CPOE/CDS systems, a discussion about an at least medium grade of recommendation might be desirable.
In general, the information gathered by this systematic review can help health care professionals, software developers, and decision makers find new avenues to optimize currently available CPOE/ CDS systems.Further more, our review provides a summary of recommended features for future CPOE/CDS systems to be developed for use in pediatric care.

Conclusion
This systematic review summarized the presently published recommended features for pediatric CPOE/CDS systems.We were able to present a standardized and graded set of features that could be useful for optimizing current CPOE/CDS systems.In addition, we see room for improvement in scientific research regarding recommendations for pediatric CPOE/CDS systems in countries outside the USA.The information gathered by this systematic review can help health care professionals, software developers, and decision makers find new starting points to optimize currently available CPOE/CDS systems.

2 by weight or m 2
NS: Not specifed (EU) in comparison to hospitals in the USA.Only one Dutch publication provided recommendations for pediatric CPOE/ CDS systems outside the USA.The varying levels of support for CPOE/CDS systems in the USA and the EU might influence the use and therefore the necessity of research about country-specific requirements.The Health Information Technology for Economic and Clinical Health (HITECH) Act was adopted in the USA in 2009.It provides incentive payments for clinicians and hospitals